She had been trying to get the AI project approved for seven months. The concept was solid. The use case was clear. The technology was available. And every time the proposal went to the finance team, it came back with the same verdict: not enough to go on.
The problem was not the idea. It was how the idea was being justified.
Most AI business cases fail at the CFO level not because the project is wrong but because the financial argument is built on the wrong foundations. Vague efficiency gains. Aspirational accuracy improvements. Competitor benchmarks that do not translate to the specific business context. These are the things that make finance teams cautious, and they are caution.
Building an AI business case that gets approved requires a different approach – one that starts with costs you can actually measure today, not benefits you are hoping to achieve tomorrow.
Start With What the Problem Costs Right Now
The most credible number in any AI business case is the cost of the current state. Not an estimate. Not an industry average. The actual, documentable cost of the process you are proposing to change, in your business, using your data.
This means calculating the fully loaded cost of the human time currently spent on the target process. Take the number of people involved, their average loaded cost per hour, the number of hours per week spent on the specific task, and annualize it. That is your baseline.
Then add the error cost. What does it cost when the manual process produces an incorrect output? Rework time. Customer impact. Compliance exposure. These are often the most significant numbers in the business case and the most frequently omitted.
Then add the capacity constraint cost. What is the business not doing because the team is occupied with this process? Revenue not pursued. Decisions delayed. Opportunities missed. This is the hardest number to calculate but often the most compelling one.
Build the Benefit Case Conservatively
Finance teams discount optimistic projections automatically. The most effective AI business cases I have seen use conservative estimates with explicit assumptions, not best-case numbers with no supporting logic.
If the AI solution is expected to reduce processing time by 70 percent, model it at 50 percent in year one and 65 percent in year two. If accuracy is expected to improve from 85 percent to 97 percent, model 90 percent. Then show the sensitivity analysis – what happens to the ROI case if the improvement is only 40 percent. If the project still returns positive value at 40 percent improvement, that is a much stronger argument than a case that only works at the best-case scenario.
The CFO reading a business case does not want to be sold. They want to be given enough honest information to make a defensible decision. Conservative estimates with clear assumptions are more persuasive than optimistic ones with none.
Include the Cost of Doing Nothing
This is the most consistently underused element of an AI business case, and often the most powerful one.
The status quo is not free. The process you are proposing to improve will continue consuming time and producing errors while the decision is delayed. Competitors who invest in this capability now will have a compounding advantage over the businesses that do not. Talent who are currently doing repetitive work that AI could handle are developing market options that do not involve staying in that role.
Quantifying the cost of inaction – in terms of continued operational cost, widening competitive gap, and retained-talent risk – often shifts the conversation from whether to invest to how quickly.
The Payback Period Is the Number That Closes
In our experience, the single metric that determines whether an AI business case gets approved is the payback period – the point at which the cumulative benefit exceeds the total investment.
For most operational AI solutions, a payback period of 12 to 18 months is well within the range finance teams will approve without significant challenge. Under 12 months is a strong case. Over 24 months requires a compelling strategic argument to accompany the financial one.
Calculate this number explicitly, show the assumptions behind it, and make it the centrepiece of the executive summary. Everything else in the document supports that number.
Where We Come In
At DoSystems, business case development is part of every AI consulting engagement. We help clients build the current-state cost model, set conservative benefit assumptions based on comparable deployments, and structure the financial argument in a form that finance teams can evaluate. The projects that get approved are the ones built on honest numbers.
#AIROI #AIBusinessCase #AIStrategy #AIConsulting #DoSystems #CFO #AIInvestment #BusinessTransformation
Comments are closed